Your AI Agent Keeps Missing The Real Bottleneck. JetBrains Rider Can Fix It Now. - The JetBrains Blog

JetBrains' Rider has introduced a new dotTrace-backed profiling skill for its AI Assistant to address the limitations of AI agents in identifying performance bottlenecks in applications. Traditionally, AI agents have struggled to accurately diagnose issues like app freezes by merely scanning code and making educated guesses, often missing the real problem. The new skill allows agents to analyze .dtp snapshots captured with dotTrace, providing precise runtime evidence to pinpoint CPU usage and inefficiencies. The implementation has significantly improved the accuracy of diagnosing performance issues, as evidenced by evaluations that showed a leap in average accuracy scores from 4.71 to 8.15 out of 10 and a doubling in perfect root-cause matches. While the enhanced accuracy comes at a slightly higher cost per run, the skill reduces time and expenses by focusing on evidence rather than speculative code analysis. This feature, available in the Rider 2026.2 EAP, bridges the gap between code analysis and runtime evidence, transforming the AI Assistant from an agent that guesses to one that knows.